Lots of non-existant networks

Abha Ahuja ahuja at merit.edu
Thu Jun 26 18:14:54 UTC 1997


Hi!

We started tracking this kind of information too... Check out:

http://www.rsng.net/pair/

We're hoping that these tools make it easier for the community to 
diagnose and troubleshoot these types of routing and policy problems.

Let us know what you think of them... 

-abha ;)


On Thu, 26 Jun 1997, Curtis Villamizar wrote:

> 
> In message <Pine.SOL.3.95.970618151012.15875A-100000 at roofdog.acsi.net>, "Eric D
> . Madison" writes:
> > There still seems to be a lot of Routes out there that are not in the
> > RADB.  I know that ANS is not excepting routes not in a routing database,
> > and others are starting to follow.  It's a good idea!  
> > 
> > I am not getting some of these routes because of the lack of an RADB
> > entry, and it's very frustrating.. 
> > 
> > Anyone else having the same problems??
> > 
> > Ohh one more thing.. Seems that there are also some routes that are being
> > advertised that don't even have a InterNIC/RIPE/APNIC entry either..
> > If they don't have an entry, then they are not real "Valid" routes, so I
> > don't want to route them.. but the Origin is from a large
> > provider..  any suggestions?  Seems that they have build their whole
> > infrastructure on these routes too.
> > 
> > Eric
> > 
> > _______________________________________________________
> >       Eric D. Madison - Senior Network Engineer -   
> >  ACSI - Advanced Data Services - ATM/IP Backbone Group  
> >    24 Hour NMC/NOC (800)291-7889 Email: noc at acsi.net
> 
> 
> The route-dumps anaylsys page has been updated.  It provides quite a
> bit more detail, some ANS specific (sorry).  The URL is:
> 
>   http://engr.ans.net/route-dumps/
> 
> New stuff on this page are:
> 
>   Listing by border AS - so we know who we can bug if a reachability
>   problem is reported and we know how much will be missing due to
>   filtering on a per border AS basis.  It also helps us concentrate on
>   routes passed to us by our direct customers.  This would be useful
>   to others if they peer with some of the same border AS.
> 
>   Reports now include (as of the most recent one):
> 
>     2477 unregistered prefixes 
>     240 prefixes unreachable due to incorrect origin AS 
>     2 prefixes with no ANS policy
>     15 prefixes with incorrect ANS policy
> 
>     2074 prefixes registered with incorrect origin AS (a warning)
> 
>   Each problem category is also sorted by border AS and origin AS
> 
>   Pages are available which report on only a specific border AS and/or
>   origin AS.  These are useful to anyone wanting to clean up their own
>   AS, since the observed paths for each prefix is provided.
> 
>   Information is available down to the prefix, with observed AS paths
>   reported, as well as the [sort of encrypted - consider it a bug] ANS
>   policy in cases where ANS policy is thought to be the problem.
> 
> Of the 22 border AS with any unregistered prefixes (46 with some sort
> of problem or warning), the worst offenders are the border AS with
> over 90 unregistered prefixes:
> 
>   AS174 (149),
>   AS286 (94),
>   AS568 (166),
>   AS701 (550),
>   AS1239 (1131),
>   AS1800 (443).
> 
> There are 398 origin AS with unregistered prefixes, most with just a
> few.
> 
> Curtis
> 
> btw- the "15 prefixes with incorrect ANS policy" both seem to be due
> to a transient where we intentionally do not accept a backup route
> from the CIX for certain providers we directly peer with yet the CIX
> was announcing one.  The "2 prefixes with no ANS policy" are one
> prefix origin AS and we addressed this after the report.
> 

__________________________________________________________________________
--------------------------------------------------------------------------
abha ahuja						ahuja at merit.edu
Merit Network, Inc.





More information about the NANOG mailing list